Solid cocoa butter has a yellowish to off-white colour. Image by Zichrini

Summary

The majority of cocoa breeding programs focus on seed yield and disease resistance, even though cocoa butte is a major commercial product. The average fat content of cocoa beans can range from roughly 45.4% to 60.3% with an average of 53.2% (taken from the Centro de Pesquisa do Cacau germplasm collection in Bahia, Brazil). The fat content appeared to be related to specific genotypes, with Upper Amazonian types having higher fat levels versus the Trinitario-Criollo and Bahian genotypes.

(add more later)

Introduction

Cacao seeds contain more fat than any other major oil crop, except for coconut which is about 64% oil (Luhs & Friedt, 1994). We know that from other oil crops such as soybean and peanut, there is a high level of diversity in their oil content, but little is known about the diversity of fat content within Theobroma cacao. Fat content in cocoa beans is important as there are increased costs associated with grinding low fat content seeds (Duncan & Veldsman, 1994; Wood & Lass, 1985). There has been a lack of attention to selecting cacao for fat content, and little is known about the effect of the environment on fat content, or the mode of inheritance for fat content.

The researchers here put together a selection criteria for a potential genetic improvement program to increase fat yield:

Characterize the available variation for fat content and its relationship with dry seed yield in a large sample of cacao germplasm.
Identify original populations with outstanding characteristics for fat content and fat yield.
Study the inheritance of fat content.
Determine the pollen effect on fat content.

Material And Methods

The collection at CEPEC (Centro de Pesquisas do Cacau) contained 949 accessions (which are distinct genetic cultivars or varieties, and often assigned a number), but only 576 were at the producing stage during this study. Open-pollinated pods were harvested from 490 of these 576 accessions. Samples included:

93 local genotypes from Southern Bahia (SIC and SIAL) and Espirito Santo (EEG)
132 CEPEC genotypes (locally selected from cultivated areas, selections from hybrid progenies, accessions of unknown origin)
70 genotypes originally collected from the Brazilian Amazon (BE, CA, CAB, CAS, CJ, CSUL, MA, RB, and others)
52 from the Peruvian Amazon (IMC, NA, PA, POUND, SCA)
37 genotypes from Costa Rica (CC and UF)
28 from Trinidad and Tobago (ICS, TSH, TSA)
23 from Mexico (RIM, P)
18 from Colombia (APA, SC, SPA, SPEC)
8 from Ecuador (EET)
9 from Venezuela (CHO, OC)
6 from Guatemala (SGU, TJ)
4 from Grenada (GS)
2 from Indonesia (DR, GW)
1 from West Africa (Amelonado)
1 from Cameroon (SNK)
1 from Samoa (LAFI)

Some of these open-pollinated pods were harvested to create three hybrid progenies in order to investigate the degree to which fat content was inherited:

POUND 12 (high fat content) x CSUL 7 (high fat content) cross
ICS 1 (low fat content) x PA 150 (high fat content) cross
SIC 4 (low fat content) x ICS 1 (low fat content) cross

Seeds were collected from either ten trees per accession for the germplasm survey, or from an individual tree for the progeny experiment. Seeds had pulp removed, dried artificially, and stored at room temperature until analyzed.

Fat was extracted according to Office International du Cacao et du Chocolat method (Anon., 1972). Yield data was collected based on the average of 5 trees for each accession over 3 years. Seed dry weight was obtained from 5 replicates of 40 seeds each, and collected once a year during the main crop season for 3 years. For each accession, a sample of 5 seeds were peeled, dried, and the weight of cotyledons and testa recorded (no replicate). Testa percent was based on zero percent moisture.

Results And Discussion

The fat content of the 490 accessions averaged 53.2%, ranging from 45.4 (CC 57 - Costa Rica) to 60.3% (NA 312 - Peru). Figure 1A illustrates the distribution of fat content among the 490 accessions. The difference among them were highly significant, with a minimum significant difference of 6.9% based on Tukey test, which indicate that there is opportunity for selecting for high fat content in cocoa. The average fat content found in this study was lower than what chocolate manufactures require which is between 56-58% fat content (Anon., 1984).

High fat content beans (above 54% fat) were collected in the Amazonian region (CSUL, SPA, NA, PA, CJ, POUND, RB). Low fat content beans (below 53%) were mostly found in Trinitario/Criollo genotypes (UF, SGU, OC, ICS, P, and CC). This trend was also observed by Wood & Lass (1985) where Amazonian hybrids were 58-60% fat content while Criollo crosses were 53% fat. Low fat content was also observed for Ecuadorian genotypes (EET) and some Bahian commercial genotypes (SIC, SIAL, EEG) which are known for having lower fat content compared to West African or Malaysian cacao (Anon., 1984; Wood & Lass, 1985).

Fat content vs Fat Yield

When it comes to actual fat yield, it was found that types with lower fat content (% of fat in the bean) (Sic, SIAL, EEG) had the highest total fat yield per plant. Fat yield per plant was calculated by:

Taking the dry seed yield (for example, CSUL = 429) and subtracting the testa (429 - 17.3% = 353.8)
Multiplying that by the fat content % (353.8 x 56.6% = 200g fat yield/plant)

See Table 1. There was a significant correlation between seed weight and testa percentage (- 0.424, p > 0.0001). However, the correlation between fat content (% fat) and fat yield was not significant. There was a small significant negative correlation between dry seed yield and fat content (Table 2).

Due to the fact that fat content does not correlate to fat yield, but that dry seed yield did correlate to fat content, means that selecting genotypes for higher cocoa butter production should be based on selecting for dry seed yield. Dry seed yield even had more variability than fat content (Pires et al., 1993). When the genotypes were analyzed by placing them into high, medium, and low fat content groups, it was observed that the highest fat content group produced less dry seed.

Dry seed yield and fat content can be considered interdependent traits (depend on each other). Selecting for higher dry seed yield (based on number of pods or pod value (number of seeds/pod x single seed weight)) would result in overall decreased fat content of the seeds. This may explain why cultivars or domesticated cacao genotypes (Criollo/Trinitario or lower Amazonian types from Bahia) often present with lower fat content values than, for instance, wild cacao types in the Upper Amazonian region.

It should be pointed out that although there was a significant negative association between fat content and dry seed yield, the correlation coefficient was rather low due to the fact that 40 genotypes were both high in fat content and dry seed yield. As well, fat content might be determined by a fewer number of genes than yield, be less affected by environmental factors, and could be tagged with molecular markers as were used in other major oil crops (Diers et al., 1992; Mansur et al., 1993).

Fat content and seed weight

Fat content was negatively associated with seed weight, and this relationship should be considered a genetic parameter of relationship (Engels, 1983) derived from genotype origin. For instance, Upper Amazonian wild genotypes have mostly smaller seeds with high fat content. In contrast, Criollo/Trinitarios and Lower Amazonian had larger seeds and lower fat content on average.

Commercial seeds from West Africa and India on the other hand had a positive correlation, with smaller seeds having lower fat content. This is likely due to pods developing during the dry season (Toxopeus & Wessel, 1970; Wood & Lass, 1985).

Crosses of high/low fat genotypes

When parent genotypes were crossed, the fat content of their progeny was similar to the parental mean (Figure 1B). Alvarado & Bullard (1961) found comparable results. Therefore, because a parent with high fat content would raise the average of the progeny, high fat genotypes should be included in hybrid testing programs.

Pollen Effect

In the diallel cross (a mating/breeding scheme used to understand link between genetics and specific traits) between 3 high fat genotypes (SPA 17, CAS 1, and CSUL 7) and 3 low fat genotypes (CC 39, ICS 9, SIC 4), a highly significant effect of pollen donor was observed, which was also observed by Jacob (1971) and Beck et al. (1977). The average increase in fat content was 2.7% (Table 4). The combining ability effects were significant, and all values above 57% fat included a high fat content parent.

Variation in fat content between seasons

Table 5 illustrates the variation in fat content between seasons in relation to genotype. The fat content of open-pollinated pods was analyzed. The genotypes were divided into two groups: high fat (above 53.5% fat content) and low fat (below 53.5% fat content). Although for the most part, genotypes stayed within their original group, even though the fat content did change. The only exceptions were SE 1 and SIC 23 which were the only two which changed groups between seasons. This strategy allowed for a clear classification of genotypes into fat content classes, which may allow for a selection criteria if selecting for superior genotypes with above average fat content.

Increasing cocoa butter yield through genetic improvement of Theobroma cacao L.